Media Planning, Buying and Advertising Technology

Introduction and Context

Performance marketing connects media investment to measurable outcomes such as clicks, leads, installations, sales, subscriptions, and revenue. Media planning determines where, when, and how often advertising should appear; media buying secures inventory at an acceptable price; advertising technology enables targeting, auction participation, delivery, measurement, optimisation, and governance.

The modern advertising marketplace is a complex, partly automated system. A campaign may involve an advertiser, agency, demand-side platform (DSP), data platform, ad exchange, supply-side platform (SSP), publisher, ad server, verification provider, and measurement partner. Performance marketers must therefore understand both commercial metrics and the technical infrastructure through which those metrics are produced.

A central managerial principle is that no pricing metric should be evaluated in isolation. A low CPM may produce poor-quality traffic, while an apparently expensive CPA may generate customers with high lifetime value. Media decisions should connect cost, delivery, conversion quality, incrementality, margin, and risk.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Distinguish CPM, CPC, CPA, CPL, CPI, and revenue-share models.
2. Calculate and interpret media efficiency metrics.
3. Explain ad auctions, quality scores, and effective ranking.
4. Describe programmatic buying and the roles of DSPs, SSPs, exchanges, and ad servers.
5. Compare open-web ecosystems with walled gardens.
6. Apply supply-path optimisation to improve efficiency and transparency.
7. Evaluate privacy, fraud, brand-safety, measurement, and governance risks.
8. Design a media-buying approach using business objectives rather than isolated platform metrics.

1. Media Planning as an Economic Decision

Media planning allocates a finite budget across audiences, channels, formats, geographies, devices, and time periods. A plan normally specifies objectives, audience definitions, reach and frequency expectations, inventory, budget, bidding strategy, measurement design, and optimisation rules.

The planning process should move through five questions:

• Business objective: Is the goal awareness, qualified demand, transactions, installations, retention, or profit?
• Conversion event: What observable action represents progress toward the objective?
• Audience and supply: Which users and media environments can generate that action?
• Economic threshold: What cost or return is acceptable?
• Evidence standard: How will the marketer distinguish correlation from incremental impact?

A useful distinction is between delivery metrics and outcome metrics. Impressions, reach, viewability, and clicks describe media delivery. Leads, purchases, gross margin, customer lifetime value, and incremental conversions describe business outcomes. Delivery metrics are necessary for diagnosis but are not automatically evidence of commercial success.

The Media Funnel

A simplified funnel moves from exposure to attention, interaction, conversion, and value. However, the funnel is not always linear. Retargeting may reach existing prospects, brand activity may increase later direct searches, and a user may encounter several channels before converting. Consequently, attribution should be treated as a measurement model rather than a literal description of consumer psychology.

2. Pricing and Performance Models

2.1 CPM: Cost per Mille

CPM is the cost of one thousand impressions. “Mille” means one thousand.

Cost is measured in currency and impressions are counted exposures. CPM is useful for comparing the price of inventory, but it does not indicate whether impressions were viewable, brand-safe, unique, or effective.

Rearrangements are:

These relationships assume that impressions and cost are measured on the same reporting basis and that no additional fees are hidden outside the stated cost.

2.2 CPC: Cost per Click

CPC is appropriate when traffic or interaction is the immediate objective. It can be distorted by accidental clicks, incentivised clicks, bots, or low-intent placements. A low CPC is valuable only if the resulting visitors are relevant and capable of progressing toward the business objective.

Click-through rate (CTR) is:

If a campaign produces 400,000 impressions and 4,000 clicks, its CTR is 1%. If cost is $8,000, CPC is $2 and CPM is $20. The three values are internally consistent because:

This relationship applies when the same impression, click, and cost definitions are used.

2.3 CPA: Cost per Acquisition or Action

“Acquisition” must be defined. It may mean a purchase, subscription, completed application, or another action. CPA is more commercially meaningful than CPC when conversion tracking is reliable, but it depends on attribution rules, conversion windows, deduplication, and the quality of the recorded action.

2.4 CPL: Cost per Lead

CPL is a specialised CPA. A lead may be a form completion, enquiry, callback request, or qualified sales opportunity. Organisations should distinguish raw leads from marketing-qualified leads and sales-qualified leads. A campaign with a higher raw CPL may be superior if it generates more qualified opportunities.

2.5 CPI: Cost per Install

CPI is common in mobile-app marketing. Install quality must be assessed through activation, retention, purchase, and fraud checks. An inexpensive install from a low-retention source may have less value than a costly install from a high-retention source.

2.6 Revenue Share and Commission Models

In a revenue-share model, the media partner receives an agreed proportion of revenue generated by referred customers.

If attributable revenue is $50,000 and the rate is 10%, payment is $5,000. The difficult issue is not multiplication but defining attributable revenue: gross or net revenue, refunds, taxes, discounts, repeat purchases, attribution window, and duplicate referrals must be specified contractually.

Advertisers should compare revenue share with contribution margin, not merely revenue. If contribution margin is 30% and the partner receives 10% of revenue, the remaining margin before other costs is 20% of revenue. A revenue-share arrangement may transfer risk from advertiser to partner, but it can also encourage aggressive attribution or low-quality referrals.

Comparative Interpretation

3. Auctions, Bids, and Quality

Most digital inventory is allocated through an auction. When an impression becomes available, eligible buyers submit bids and associated campaign information. The platform ranks eligible ads and selects a winner according to its auction rules, policy constraints, predicted user value, and commercial considerations.

A simplified ranking concept is:

This is a conceptual model, not a universal platform formula. Actual systems may use predicted action rates, advertiser relevance, landing-page experience, reserve prices, contextual signals, user controls, and other constraints. Quality scores are generally diagnostic or ranking-related signals rather than universal numeric measures that can be directly compared across platforms.

In a second-price-like auction, the winner may pay an amount related to the next-highest eligible bid rather than the full submitted bid. In first-price auctions, the winner generally pays its submitted clearing bid, subject to the platform’s rules. Modern exchanges may use variations, floors, deal terms, and multiple sequential auctions, so practitioners must consult the relevant platform specification.

Worked Auction Example

Assume three eligible bids for one impression:

Under the illustrative ranking rule, Buyer B wins because 3.84 is the highest rank, despite not submitting the highest bid. This example demonstrates why relevance, predicted engagement, landing-page experience, and policy compliance can affect delivery. It does not imply that every platform uses this exact multiplication rule.

A performance marketer should avoid treating quality optimisation as a method of simply lowering price. Improving creative relevance and post-click experience may increase conversion probability, improve delivery, and lower effective acquisition cost, but results must be tested rather than assumed.

4. Programmatic Advertising Architecture

Programmatic advertising is the automated buying and selling of digital advertising inventory. Automation does not mean that every transaction is open auction inventory; programmatic direct, private marketplaces, preferred deals, and guaranteed arrangements are also programmatic.

Figure: Simplified programmatic transaction flow; real implementations may contain additional intermediaries and parallel data services.

Demand-Side Platforms

A DSP enables advertisers or agencies to manage bids, budgets, audiences, frequency controls, creatives, optimisation, and reporting across available inventory. It may connect to multiple exchanges and SSPs. DSP algorithms estimate the probability and value of an outcome, subject to budget pacing and eligibility constraints.

Supply-Side Platforms

An SSP helps publishers package, price, and sell inventory. It manages demand connections, floor prices, deal terms, yield controls, and reporting. SSPs may also support identity, consent, inventory quality, and supply-chain declarations.

Ad Exchanges

An exchange facilitates transactions between demand and supply. It may conduct auctions, transmit bid requests and responses, apply policy rules, and provide transaction records. The boundary between exchanges and SSPs is not always commercially or technically identical across vendors.

Ad Servers

An ad server stores or references creatives, makes delivery decisions, applies frequency and sequencing rules, records impressions and clicks, and supports trafficking and reporting. Advertiser ad servers and publisher ad servers may both participate in a transaction. Discrepancies can arise from different counting rules, time zones, caching, fraud filtering, or tracking restrictions.

Data Platforms

A customer data platform (CDP) generally unifies first-party customer data for known organisational uses. A data management platform (DMP) traditionally organised audience segments, often for advertising activation, frequently using pseudonymous identifiers. A clean room enables controlled matching or analysis between parties without directly exposing underlying records. These categories overlap in practice, so governance and actual data flows matter more than product labels.

5. Walled Gardens and Open-Web Buying

A walled garden is a platform that controls substantial parts of identity, inventory, auction, measurement, and user experience. Examples include major search, social, retail-media, and app ecosystems. Walled gardens can offer rich first-party signals and large audiences, but advertisers may face limited user-level transparency, platform-specific attribution, restricted experimentation, and dependence on the platform’s reporting definitions.

Open-web buying can provide broader publisher access, independent verification, and greater supply-path choice. It may also involve more intermediaries, fragmented identity, variable inventory quality, and greater exposure to fraud or brand-safety risk.

A sound media strategy does not assume that either environment is inherently superior. It compares incremental reach, qualified outcomes, marginal cost, measurement reliability, privacy compliance, and strategic dependence.

6. Supply-Path Optimisation

Supply-path optimisation (SPO) is the systematic evaluation and selection of efficient, transparent, and quality-assured routes from advertiser demand to publisher inventory. The same impression may be reachable through several SSPs or resellers. Multiple routes can create duplicated fees, auction conflicts, latency, and limited accountability.

An SPO review should examine:

• total cost, including technology fees and media working cost;
• win rate, clearing price, latency, and timeout rate;
• duplicate bid opportunities and auction mechanics;
• publisher and app-ads.txt or sellers.json declarations where applicable;
• supply-chain object accuracy and authorised seller status;
• viewability, invalid traffic, brand suitability, and contextual quality;
• identity, consent, data use, and retention practices;
• unique reach and incremental value rather than superficial platform volume.

The objective is not to reduce the number of vendors automatically. A route may be valuable because it provides unique inventory, a private deal, regional expertise, or stronger quality controls. The correct question is whether its incremental value justifies its cost and risk.

7. Measurement, Attribution, and Optimisation

Attribution assigns credit for a conversion according to a specified rule. Common rules include last click, first click, linear, position-based, and algorithmic attribution. Each answers a different question and may produce different budget recommendations.

Attribution is not the same as incrementality. Incrementality asks whether the conversion would have occurred without the advertising exposure. The strongest practical designs include holdout experiments, geo experiments, conversion lift studies, and carefully controlled time-series approaches. Platform-reported conversions are useful operational signals but should not be treated as unbiased causal estimates by default.

A basic return-on-ad-spend measure is:

ROAS is a ratio, not a profit measure. If revenue is $120,000 and media cost is $30,000, ROAS is 4.0, or 400%. Profitability additionally requires product costs, fulfilment, commissions, technology, agency fees, refunds, and overhead. A margin-aware measure may use contribution profit in the numerator instead of revenue.

For customer acquisition, a useful economic comparison is:

This inequality assumes that expected contribution value is measured over a defined period and includes expected retention, repeat purchasing, variable costs, and uncertainty. It should not be interpreted as permission to spend the entire expected value when cash flow, risk, and payback requirements impose stricter limits.

8. Applied Case: Illustrative Mobile Subscription Campaign

Background

A fictional streaming application plans to spend $100,000 over four weeks. The objective is to acquire paying subscribers, not merely installs. The team can use social video, search, and programmatic mobile inventory.

Problem

Initial platform reports show that programmatic inventory has the lowest CPI, while search has the highest CPI. Management is considering moving most budget to programmatic.

Analysis

Programmatic appears efficient on CPI but inefficient on paid-subscriber CPA. Search produces fewer installs yet more subscribers than programmatic. The team then examines retention and contribution value. Suppose the expected 12-month contribution per subscriber is $70 for search, $60 for social, and $55 for programmatic. On a simple contribution-minus-media basis, channel-level results are:

• Social: 700 × $60 − $35,000 = $7,000.
• Search: 800 × $70 − $30,000 = $26,000.
• Programmatic: 400 × $55 − $35,000 = −$13,000.

Approach and Solution

The organisation does not eliminate programmatic immediately. It first audits placement quality, app-identifier transparency, click-to-install patterns, post-install events, and invalid traffic. It introduces a qualified-subscription optimisation event, excludes suspicious supply, applies frequency controls, and tests a private marketplace route. Search receives additional budget subject to marginal CPA monitoring; social is retained while creative and onboarding are improved.

Outcome

The case illustrates why buying against CPI alone would have misallocated budget. The decision metric should reflect the business objective and downstream value. Results remain illustrative: a real decision would require confidence intervals, incrementality evidence, cash-flow constraints, and marginal rather than average performance.

Learning Points

1. The denominator defines what a metric can and cannot tell the manager.
2. Cheap inventory may produce low-quality or fraudulent outcomes.
3. Post-install or post-lead quality is essential to performance evaluation.
4. Optimisation events should be close enough to value to guide algorithms, but frequent enough to provide adequate learning volume.
5. Vendor reports require independent validation and governance.

9. Governance, Ethics, and Contemporary Developments

Advertising technology processes behavioural, transactional, device, and sometimes sensitive data. Responsible practice requires a lawful basis for processing, meaningful consent where required, purpose limitation, data minimisation, security, retention controls, access rights, and transparent disclosures. Jurisdictional requirements differ, so legal review is essential.

Important risks include discriminatory targeting, exclusion from housing or employment opportunities, opaque automated decisions, manipulative creative, excessive frequency, children’s privacy, location misuse, and surveillance beyond reasonable user expectations. Frequency capping and exclusion lists are not merely efficiency tools; they can reduce consumer harm.

Invalid traffic includes bots, manipulated impressions, fake clicks, install farms, domain spoofing, and other non-genuine activity. Brand safety concerns adjacency to harmful or unsuitable content, while brand suitability concerns whether a particular context fits the brand’s risk tolerance. Independent verification, contractual audit rights, authorised-seller checks, and anomaly monitoring are important controls.

Artificial intelligence is increasingly used for bidding, creative variation, forecasting, audience modelling, and fraud detection. Its use creates risks of feedback loops, opaque optimisation, biased outcomes, synthetic creative errors, and inappropriate use of personal data. Human oversight should define objectives, constraints, escalation rules, testing procedures, and stop conditions. Automated optimisation should not be allowed to redefine the organisation’s ethical or commercial objective silently.

The industry is also moving toward reduced dependence on third-party identifiers, greater use of first-party data, contextual signals, consented clean-room analysis, server-side measurement, and privacy-enhancing technologies. These developments may improve privacy but can also reduce independent visibility. Measurement design must therefore combine platform data with experiments, aggregated analysis, and carefully documented assumptions.

Best-Practice Framework

A professional media-buying operation should:

1. Define the business outcome and economic threshold before selecting a buying model.
2. Maintain a metric dictionary specifying numerator, denominator, time zone, attribution window, and exclusions.
3. Separate platform optimisation metrics from executive success metrics.
4. Use independent verification for viewability, invalid traffic, and brand suitability.
5. Audit fees, intermediaries, sellers, data use, and supply paths.
6. Test incrementality rather than relying exclusively on attributed conversions.
7. Monitor marginal performance, not just campaign averages.
8. Document consent, retention, access, and algorithmic governance.
9. Reconcile ad-server, platform, analytics, and financial records.
10. Treat automation as a controlled decision system requiring human accountability.

Chapter Summary

Media planning and buying are economic and technological disciplines. CPM measures the price of impressions; CPC measures clicks; CPA, CPL, and CPI measure specified actions; and revenue-share models compensate partners according to attributable revenue. Each model is useful only when its denominator is clearly defined and connected to business value.

Digital advertising auctions rank eligible opportunities using bids, quality, predicted outcomes, and platform rules. Programmatic ecosystems link advertisers, DSPs, exchanges, SSPs, publishers, ad servers, verification providers, and data platforms. Walled gardens offer scale and first-party signals but may limit transparency; open-web buying offers route choice but introduces fragmentation and quality risks. Supply-path optimisation evaluates the cost, quality, transparency, and incremental value of routes to inventory.

Effective management requires more than platform-reported conversions. Attribution should be distinguished from incrementality, ROAS from profit, and low acquisition cost from high customer value. Privacy, fraud, brand safety, fairness, and algorithmic accountability are central professional responsibilities. The strongest practice combines clear economics, controlled experimentation, technical literacy, independent validation, and human governance.

Key Terms

CPM: Cost per one thousand advertising impressions.

CPC: Cost per click.

CPA: Cost per specified acquisition or action.

CPL: Cost per lead.

CPI: Cost per app install.

Revenue share: Compensation based on an agreed proportion of attributable revenue.

CTR: Click-through rate, calculated as clicks divided by impressions.

Ad auction: A mechanism for selecting and pricing eligible advertising opportunities.

Quality score: A platform-specific relevance or quality signal that may affect ranking or delivery.

Programmatic advertising: Automated buying and selling of digital advertising inventory.

DSP: Demand-side platform used by advertisers to buy inventory.

SSP: Supply-side platform used by publishers to sell inventory.

Ad exchange: Marketplace facilitating automated transactions between demand and supply.

Ad server: Technology that manages creative delivery, counting, trafficking, and reporting.

CDP: Customer data platform supporting governed first-party customer-data activation.

DMP: Data management platform traditionally used to organise advertising audience segments.

Walled garden: Platform controlling substantial inventory, identity, data, and measurement functions.

Supply-path optimisation: Evaluation and selection of efficient and transparent routes to inventory.

Incrementality: The causal effect attributable to advertising exposure.

Viewability: The extent to which an ad was capable of being seen under a defined standard.

Review Questions

Short Answer Questions

1. What is the difference between CPM and CPC?
2. Why can a low CPI be misleading in mobile marketing?
3. What functions does a DSP perform?
4. Define supply-path optimisation.
5. Why is ROAS not equivalent to profit?

Long Answer Questions

1. Compare CPM, CPC, CPA, CPL, CPI, and revenue-share models, including the contexts in which each is most appropriate.
2. Explain how an advertising auction can select a lower bidder when quality or relevance is included in ranking.
3. Describe the roles and relationships of DSPs, SSPs, exchanges, ad servers, and data platforms.
4. Evaluate the advantages and limitations of walled gardens compared with open-web programmatic buying.
5. Discuss the ethical, privacy, fraud, and governance responsibilities of a performance-marketing manager.

Analytical / Application Questions

1. A campaign spends $24,000 and generates 600 qualified leads, of which 120 become customers. Calculate CPL and customer CPA, then explain which metric should guide budget decisions if customer quality varies by channel.
2. A retailer reports high ROAS from retargeting but has no holdout group. Design an experiment to assess whether the campaign generated incremental sales.
3. A mobile advertiser has a low-cost supply path with high discrepancy, poor app transparency, and weak retention, and a more expensive path with verified publishers and better retention. Recommend a decision framework for choosing between them.

Multiple Choice Questions with Answers

Question 1

A campaign spends $12,000 and delivers 600,000 impressions. What is its CPM?

A. $2

B. $20

C. $50

D. $200

Correct Answer: B. $20

Question 2

Which metric measures media cost divided by completed leads?

A. CPL

B. CPM

C. CPI

D. CTR

Correct Answer: A. CPL

Question 3

Under the illustrative rule Ad rank = Bid × Quality factor, which buyer wins: A bids $4 with quality 0.8; B bids $3.2 with quality 1.2; C bids $4.5 with quality 0.6?

A. A

B. B

C. C

D. The highest bid always wins

Correct Answer: B. B

Question 4

Which technology primarily enables an advertiser to manage bids and audience buying across inventory sources?

A. SSP

B. DSP

C. Publisher ad server

D. Payment gateway

Correct Answer: B. DSP

Question 5

What is the principal objective of supply-path optimisation?

A. Maximise the number of intermediaries

B. Select efficient, transparent, quality-assured routes to inventory

C. Replace all conversion tracking with clicks

D. Guarantee that every auction uses first-price pricing

Correct Answer: B. Select efficient, transparent, quality-assured routes to inventory

Question 6

A campaign spends $30,000 and generates 800 conversions. What is the CPA?

A. $26.67

B. $37.50

C. $40.00

D. $240.00

Correct Answer: B. $37.50

Question 7

Which statement best distinguishes attribution from incrementality?

A. Attribution assigns credit; incrementality estimates causal additional impact

B. Attribution always proves causality; incrementality counts impressions

C. They are identical terms

D. Incrementality is only a billing metric

Correct Answer: A. Attribution assigns credit; incrementality estimates causal additional impact

Question 8

What is a characteristic of a walled garden?

A. It has no control over identity or inventory

B. It controls substantial platform data, inventory, or measurement functions

C. It can only sell print advertising

D. It prohibits automated buying

Correct Answer: B. It controls substantial platform data, inventory, or measurement functions

Question 9

If attributable revenue is $50,000 and the agreed revenue-share rate is 10%, what is the partner payment?

A. $500

B. $5,000

C. $10,000

D. $50,000

Correct Answer: B. $5,000

Question 10

Which is the strongest reason to evaluate post-install behaviour in an app campaign?

A. Installs always equal revenue

B. It reveals whether acquired users activate, retain, or monetise

C. It removes the need for fraud controls

D. It guarantees lower CPM

Correct Answer: B. It reveals whether acquired users activate, retain, or monetise